Simply put, quantitative stock selection is to establish a model with quantitative methods, and then select a stock portfolio, hoping that the stock portfolio can obtain an investment method that exceeds the benchmark rate of return.

What are the risk characteristics of quantitative stock selection?

We take two typical strategies that use quantitative stock selection methods in the market as examples:

First, the market-neutral strategy.

For the market-neutral strategy, the main goal is to choose a high alpha stock portfolio by quantifying stock selection and hedging stock index futures. In this way, the market risk of the stock portfolio can be stripped and the pure alpha income can be harvested. Therefore, a standard pure market-neutral strategy should be less affected by market fluctuations, and then obtain good excess returns steadily. Therefore, in general, the neutral strategy is a relatively stable investment strategy, with less retracement risk and stable fluctuation, and the maximum retracement is smaller than that of pure bull products.

Second, the exponential enhancement strategy.

At present, the mainstream index enhancement strategy in the market is mainly evolved from the original market neutral strategy. In order to improve the efficiency of capital use and strive for higher returns, the hedging part of stock index futures in market neutral strategy is removed, and a pure long portfolio of stocks is directly constructed. Quantitative stock selection method is used to select a basket of stocks, track the index and control the tracking error. The purpose is to obtain the income higher than the market index under the premise of taking market risks, not only the pure alpha income provided by neutral strategy, but also the income brought by the market itself.

At present, there are two kinds of index enhancement products: CSI 300 Index Enhancement and CSI 500 Index Enhancement. There are relatively more products that track the CSI 500 Index. Because of the cancellation of stock index hedging, the products with enhanced index completely exposed the market risk, thus striving for higher returns. Therefore, index-enhanced products have the characteristics of high risk and high return. Under normal circumstances, it will rise and fall together with the fluctuation of the index tracked by the product, but it will generally rise higher than the index when it rises, and lose less than the index when it falls. Although the overall strategy fluctuates greatly, there may be a big retreat during the investment period. However, because index-enhanced products have higher efficiency of capital use and stronger compound interest effect than pure neutral products, it is easier to get higher returns than neutral products under the condition of little risk in the market.

The most common quantitative stock selection model

Generally speaking, the mainstream quantitative stock selection strategies in the market can be divided into two categories: the first category is fundamental stock selection, and the second category is market behavior stock selection. Among them, the fundamental stock selection models mainly include: multi-factor model, style rotation model and industry rotation model. The stock selection models of market behavior mainly include cash flow model, momentum reversal model, consensus expectation model, trend tracking model and chip stock selection model.

Institutions engaged in quantitative investment in the market use various quantitative stock selection models to build stock portfolios. With the help of modern statistics and mathematical methods, they look for various "high probability" strategies and laws that can bring stable returns to the portfolio from massive historical big data. On this basis, it is summarized into factors and model programs, and finally independent investment is made in strict accordance with these quantitative model combinations. Among many stock selection models, multi-factor stock selection model is one of the most widely used models in quantitative stock selection institutions. The basic principle of multi-factor model is to adopt a series of factors as stock selection criteria, and stocks that meet these factors are bought and those that do not meet them are sold.

The core principle of multi-factor model is to find those factors that are most relevant to the enterprise's rate of return. There are two main differences in the core of various multi-factor models. The first is that the selected factors may be different, and the second is that the combination of factors and the distribution of weights will be different. Combining these two points will lead to different stock combinations finally selected by different institutions. Generally speaking, the specific stock selection methods of multi-factor stock selection model are divided into scoring method and regression method.

Scoring method is to score stocks according to the size of each factor, then get a total score according to a certain weight, and then screen stocks according to the total score.

Regression method is to use past stock returns to regress multiple factors to get a regression equation, and then substitute the latest factor values into the regression equation to get a prediction of future stock returns, and then select stocks on this basis.